Monitoring the Ichkeul Marches Vegetations (North Tunisia): the Use of Object-Oriented Multi-Resolution Image Classification and the Combined Topographic-Bathymetric Model

摘要

The purpose of this study is to show the feasibility of the integrated topographic-bathymetric data to improve the object-oriented classification of a multi-spectral imagery in order to monitor the coastal wetland vegetation. The study area concerns the Ichkeul Park : lake, marshes and mountain located in the north of Tunisia. First we settle the integrated topographic-bathymetric digital terrain model (DTM) by combining three datasets: the lake bathymetry, the marshes and the mountain topographies. Then the new topographic-bathymetric DTM was used with multi-spectral imagery from the MSS Landsat (1972), TM Landsat (1987), ETM+ Landsat (2001) and Aster (2007) in order to classify Ichkeul marshes. This is done by using the object-oriented classification based on botanic field observations. This method improved global precision of marshes vegetation mapping. It gives a 92% global classification precision with more than nine communities in 2007. However an Aster supervised classification based on pixel approach allowed identifying four plant associations with a global precision of 82% (Ghrabi and al. 2006). We provide herein therefore an operational tool to monitor any changes in wetland areas.